Human forecasting accuracy improves through the “wisdom of the crowd” effect, in which aggregated predictions tend to outperform individual ones. Past research suggests that individual large language models (LLMs) tend to underperform compared to human crowd aggregates. We simulate a wisdom of the crowd effect with LLMs. Specifically, we use an ensemble of 12 LLMs to make probabilistic predictions about 31 binary questions, comparing them with those made by 925 human forecasters in a 3-month tournament. We show that the LLM crowd outperforms a no-information benchmark and is statistically indistinguishable from the human crowd. We also observe human-like biases, such as the acquiescence bias. In another study, we find that LLM predictions (of GPT-4 and Claude 2) improve when exposed to the median human prediction, increasing accuracy by 17 to 28%. However, simply averaging human and machine forecasts yields more accurate results. Our findings suggest that LLM predictions can rival the human crowd’s forecasting accuracy through simple aggregation.
Changes in personality traits in patients with Alzheimer's disease (AD) are extremely common throughout the course of the pathology, and these behavioral changes present themselves as challenges in clinical management and as a significant cause of caregivers’ burden. Objective: Using a personality inventory based on the five-factor model of personality, this study aimed to assesses the change in these factors by comparing the premorbid and current personality of individuals recently diagnosed with AD. Methods: A total of 30 AD patients were recruited, and their respective family members responded to the personality inventory at home through a hosted site. The patients were also divided into two groups according to the Clinical Dementia Rating (CDR): mild dementia (CDR 1) and moderate dementia (CDR 2). Results: Among all patients, there was a significant increase in neuroticism factor levels and a significant decrease in the extraversion, conscientiousness, openness, and socialization factors. When comparing the groups, only the extraversion factor showed a difference, with CDR 1 group accusing a higher change in scores. Higher scores in the factor neuroticism in the premorbid personality correlated with the current severity of the disease. Conclusions: This research draws the attention of family members and health professionals to changes in personality traits or behavior of relatives or patients, because it can reflect an underlying neurodegenerative process.
Item Response Theory represents one of the major advances in the field of developing valid and reliable measures in psychology. Among the main models used in this perspective are the Rasch model and the logistic models. These parametric models, however, are not suitable for all applications in psychology, since a substantial number of databases in psychology do not satisfy the assumptions of these models: unidimensionality; latent monotonicity; local independence; and, for some models, non-intersecting functions. Given this framework, the objective of this study was to present the theoretical and practical foundations of Mokken Scale Analysis (MSA). We present some historical issues involving the development of MSA, in addition to the main characteristics and assumptions of the two models used in this perspective. After exemplifying a MSA application, limitations and final considerations are presented, supporting the decision-making process for researchers who come to use MSA.
Background: Negative body image and disordered eating are common among adolescents and young adults. The Sociocultural Attitudes Towards Appearance Questionnaire-4-Revised (SATAQ-4R) captures the internalization of societal appearance ideals and perceived pressures from others but has not been validated in a Norwegian adolescent population. Methods: The current study explored the factor structure of SATAQ-4R in a sample of adolescent Norwegian males and females (n = 1558, mean age 17.04 ± 0.95) using confirmatory factor analysis (CFA) for a 6- and 7-factor structure in females, and a 7-factor structure in males. Correlations between subscales, internal consistency and reliability, and comparisons with convergent measures (disordered eating, body mass index, negative influence of social media) were explored. Results: The CFA supported a 7-factor structure of the SATAQ-4R for both males and females. Internal consistency and reliability were acceptable. SATAQ-4R subscales largely correlated with disordered eating and additional convergent measures. Conclusions: Results confirm the structure and reliability of the SATAQ-4R in a Norwegian adolescent population. The subscales showed good convergent validity, and high scores on internalization and societal pressures were related to higher levels of disordered eating and negative social media influence. The Norwegian version of the SATAQ-4R thus demonstrates good psychometric properties in adolescent males and females, and is well suited to capture internalization and sociocultural pressures that particularly affect adolescents. Results highlight the need to continue working towards reducing adverse internalization and improving body image among adolescents.